Abstract
Purpose :
Our purpose was to develop and validate the automated speckle tracking algorithm for choroidal sclera interface (CSI) to quantify the pulsatile change of choroidal thickness with each heartbeat using high-speed optical coherence tomography (OCT) that incorporates time series.
Methods :
The segmentation of CSI at the reference frame was implemented based on graph node search on flattened B-scan with respect to the posterior retinal pigment epithelium (RPE) (Mazzaferri et al, Sci Rep, 2017). The identified nodes for CSI were then tracked in the successive frames using a correlation-based speckle tracking algorithm. The OCT image at each frame was divided into kernels (13*13 pixels) and each kernel represents a unique speckle pattern. Maximum cross-correlation coefficient in a designated search window indicates the highest similarity of the speckle pattern, thus marks the new location of the kernel at the subsequent frame. To validate the tracking algorithm, the image matrix was translated by a uniform amount of 1, 2, and 3 pixels in the axial direction (the through-thickness direction). The displacement of CSI nodes was calculated using the tracking algorithm and compared with the intended displacement. The average axial displacement of all CSI nodes was plotted as the cumulative choroidal thickness change curve, which was then filtered to remove noise using a band-pass filter (0.6-1.4 * heart rate).
Results :
The average calculated displacements of CSI nodes based on speckle tracking of the OCT image were 0.987±0.018, 1.994±0.014 and 2.992±0.015 pixels for 1, 2 and 3 pixels uniform translation, respectively. The corresponding errors were 1.27%, 0.27% and 0.27%. Fig1A shows the tracked location of CSI nodes at the diastolic (red) and systolic (green) pressure. The frames of diastolic and systolic pressures were identified at the trough and peak of one selected cycle in the filtered curve of cumulative choroidal thickness change as shown in Fig1B. The calculated choroidal thickness change is 2.31 pixels (8.95µm).
Conclusions :
We demonstrated the feasibility of OCT image-based speckle tracking algorithm for estimation of pulsatile change of choroidal thickness. This technique may offer a useful tool for clinical biomechanical evaluation, such as calculation of ocular rigidity.
This is a 2021 Imaging in the Eye Conference abstract.